Activity and Thermostability of GH5 Endoglucanase Chimerasfrom Mesophilic and Thermophilic Parents

Fei Zheng,a,b Josh V. Vermaas,c Jie Zheng,a Yuan Wang,a Tao Tu,a Xiaoyu Wang,a,b Xiangming Xie,b Bin Yao,a

Gregg T. Beckham,d Huiying Luoa

aKey Laboratory for Feed Biotechnology of the Ministry of Agriculture, Feed Research Institute, Chinese Academy of Agricultural Sciences, Beijing, People’s Republic ofChina

bCollege of Biological Sciences and Biotechnology, Beijing Forestry University, Beijing, People’s Republic of ChinacBiosciences Center, National Renewable Energy Laboratory, Golden, Colorado, USAdNational Bioenergy Center, National Renewable Energy Laboratory, Golden, Colorado, USA

ABSTRACT Cellulases from glycoside hydrolase family 5 (GH5) are key endogluca-nase enzymes in the degradation of diverse polysaccharide substrates and are usedin industrial enzyme cocktails to break down biomass. The GH5 family shares a ca-nonical (��)8-barrel structure, where each (��) module is essential for the enzyme’sstability and activity. Despite their shared topology, the thermostability of GH5 en-doglucanase enzymes can vary significantly, and highly thermostable variants are of-ten sought for industrial applications. Based on the previously characterized thermo-philic GH5 endoglucanase Egl5A from Talaromyces emersonii (TeEgl5A), which has anoptimal temperature of 90°C, we created 10 hybrid enzymes with elements of themesophilic endoglucanase Cel5 from Stegonsporium opalus (SoCel5) to determinewhich elements are responsible for enhanced thermostability. Five of the expressedhybrid enzymes exhibit enzyme activity. Two of these hybrids exhibited pronouncedincreases in the temperature optimum (10 and 20°C), the temperature at which theprotein lost 50% of its activity (T50) (15 and 19°C), and the melting temperature (Tm)(16.5 and 22.9°C) and extended half-lives (t1/2) (�240- and 650-fold at 55°C) relativeto the values for the mesophilic parent enzyme and demonstrated improved cata-lytic efficiency on selected substrates. The successful hybridization strategies werevalidated experimentally in another GH5 endoglucanase, Cel5 from Aspergillus niger(AnCel5), which demonstrated a similar increase in thermostability. Based on molecu-lar dynamics (MD) simulations of both the SoCel5 and TeEgl5A parent enzymes andtheir hybrids, we hypothesize that improved hydrophobic packing of the interfacebetween �2 and �3 is the primary mechanism by which the hybrid enzymes increasetheir thermostability relative to that of the mesophilic parent SoCel5.

IMPORTANCE Thermal stability is an essential property of enzymes in many in-dustrial biotechnological applications, as high temperatures improve bioreactorthroughput. Many protein engineering approaches, such as rational design anddirected evolution, have been employed to improve the thermal properties ofmesophilic enzymes. Structure-based recombination has also been used to fuseTIM barrel fragments, and even fragments from unrelated folds, to generate newstructures. However, little research has been done on GH5 endoglucanases. Inthis study, two GH5 endoglucanases exhibiting TIM barrel structure, SoCel5 andTeEgl5A, with different thermal properties, were hybridized to study the roles ofdifferent (��) motifs. This work illustrates the role that structure-guided recombi-nation can play in helping to identify sequence function relationships withinGH5 enzymes by supplementing natural diversity with synthetic diversity.

KEYWORDS (��)8-barrel structure, GH5 endoglucanase, hybrid enzymes,structure-based recombination, thermostability

Citation Zheng F, Vermaas JV, Zheng J, WangY, Tu T, Wang X, Xie X, Yao B, Beckham GT, LuoH. 2019. Activity and thermostability of GH5endoglucanase chimeras from mesophilic andthermophilic parents. Appl Environ Microbiol85:e02079-18. https://doi.org/10.1128/AEM.02079-18.

Editor Haruyuki Atomi, Kyoto University

Copyright © 2019 American Society forMicrobiology. All Rights Reserved.

Address correspondence to Gregg T. Beckham,Gregg.Beckham@nrel.gov,or Huiying Luo, luohuiying@caas.cn.

F.Z. and J.V.V. contributed equally to this work.

Received 27 August 2018Accepted 4 December 2018

Accepted manuscript posted online 14December 2018Published

ENZYMOLOGY AND PROTEIN ENGINEERING

crossm

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 1Applied and Environmental Microbiology

20 February 2019

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

Cellulases are a class of industrially important enzymes that have been widely usedfor biotechnological applications (1, 2). A subset of cellulases isolated from ther-

mophilic microbes are highly thermostable and display higher cellulolytic activity andlonger half-life at elevated temperatures, which can in turn improve the economicviability of industrial processes by increasing enzymatic hydrolysis rates via operationat higher temperature (3, 4). Members of the glycoside hydrolase family 5 (GH5)endoglucanases are common components of enzyme cocktails for biomass conversion,creating new glycan chain ends for processive cellulases to depolymerize. Representa-tives across the GH5 family are able to act on diverse oligosaccharide substrates (5). Theendoglucanase Egl5A from Talaromyces emersonii (TeEgl5A) exhibits the highest ther-mostability known so far for this family, with an optimal temperature of 90°C (6).However, this is an outlier among fungal GH5 enzymes, with the best-characterizedGH5 enzymes, including EglA from Piromyces rhizinflata (PrEglA) (7), Cel5A from Ther-moascus aurantiacus (TaCel5A) (8), Cel5A from Hypocrea jecorina (Trichoderma reesei)(TrCel5A) (9), Cel5A from Ganoderma lucidum (GlCel5A) (10), and Cel5 from Aspergillusniger (AnCel5) (11), exhibiting optimal temperatures below 70°C. These enzymes areless well suited for some industrial applications, so engineering a thermostable fungalendoglucanase that retains high catalytic activity is desirable.

Various protein engineering approaches, such as rational design and directedevolution, have been employed to improve the thermal properties of mesophilic fungalcellulases (12–14). Among these, SCHEMA, a computational approach to select blocksof sequence with minimal disruption of residue-residue contacts in the resultingfunctional hybrids (15), offers a useful tool to improve enzyme thermostability. Thisrecombination method has been used to generate novel recombinant enzymes of�-lactamase, �-glucosidases, and GH6 chimeras with significantly higher levels ofactivity and thermostability (16–18). Structure-based fusion of subdomains belongingto different proteins is also an effective method for creating hybrid enzymes with newproperties (19).

GH5 enzymes consist of an 8-fold repeat of (��) units that form a barrel, a common(��)8 fold that prior experiments have shown is amenable to improvement through thecombinatorial shuffling of polypeptide segments to improve or add functionalities tothe protein. For example, the N- and C-terminal four (��)4-barrels of histidine-biosynthetic enzymes were assembled to give two highly active hybrid enzymes, HisAFand HisFA, with the (��)8 fold (20, 21). A high degree of internal structure was exhibitedin these two fusion proteins. Sequence-function analysis showed that the recombinedprotein fragments contributed additively to enzymatic properties in a given chimera(22). In a subsequent experiment, the half-barrel of HisF was replaced with the(��)5-flavodoxin-like fold from the unrelated but structurally compatible bacterialresponse regulator CheY, resulting in a stable protein with a (��)8-like fold (23). Later,a catalytically active form of the symmetrical barrel was obtained by fusing two copiesof the C-terminal half-barrel HisF-C of HisF (24). Together, these examples suggest thatsymmetrical (��)8-barrels like those of GH5 enzymes are a suitable scaffold for engi-neering new enzyme properties or functionalities (25).

In this study, we use the plasticity of the classical (��)8-barrel fold highlighted aboveto probe the relationship between structure and thermostability in fungal GH5 endo-glucanases by conducting structure-guided protein engineering. The endoglucanaseEgl5A from T. emersonii (TeEgl5A) (6) exhibits high thermostability, retaining almost allof its activity after incubation at 70°C for 1 h. In contrast, the endoglucanase Cel5 fromStegonsporium opalus (SoCel5) that we are investigating is a typical mesophilic enzyme,retaining �20% activity at 70°C for 10 min. SoCel5 shares the (��)8-barrel structure withTeEgl5A, along with 51% sequence identity, but shows a much lower temperatureoptimum, 60°C compared to 90°C. By using the fusion approach, the combinations ofthe first four (��) module(s) of TeEgl5A were introduced into SoCel5, producing 10hybrid enzymes to probe the relationship between structure and thermostability (Table1; Fig. S1 in the supplemental material).

Two hybrid enzymes exhibit substantial improvements in thermostability relative to

Zheng et al. Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 2

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

the mesophilic parent and catalytic efficiency for specific substrates, which have thepotential to lower process costs in industrial bioconversion processes. The functionalroles of this N-terminal sequence were also verified in another GH5 endoglucanasefrom Aspergillus niger (11) and its hybrid. With this work, we determined the structuralregions in TeEgl5A that contribute to its high thermostability. Comparative moleculardynamics (MD) simulations suggest that improved hydrophobic packing of the inter-face between �2 and �3 helices is the primary mechanism behind the improvedthermostability of the hybrid. These simulations also indicate a TeEgl5A-specific hydro-gen bond network surrounding R52 that may be an attractive target to further improveGH5 thermostability.

RESULTSCloning and sequence analysis of SoCel5. A gene fragment, 341 bp in length, was

amplified from the genomic DNA of S. opalus, using GH5-specific primers. The 5= and3= flanking regions were obtained by thermal asymmetric interlaced PCR (TAIL-PCR) andassembled with the known sequence to give full-length SoCel5 (1,062 bp). Sequenceanalysis indicated that the open reading frame (ORF) of SoCel5 is interrupted by oneintron (69 bp). The cDNA of SoCel5 contained 993 bp that encoded a GH5 endogluca-nase of 330 amino acids with an estimated molecular mass of 35 kDa and a predictedpI of 4.94. Based on high sequence homology with other GH5 endoglucanases ofknown structure, SoCel5 likely contains only one catalytic domain with a (��)8-barrelfold. Similarly, the N-terminal 19 amino acids were predicted to be a signal peptide, andthree N-linked glycosylation sites (Asn23, Asn64, and Asn76) are possible, based onprotein sequence and structure (Fig. 1).

Design and production of a chimeric enzyme library. The thermophilic TeEgl5Aenzyme (temperature optimum at 90°C) exhibits high catalytic efficiency and broadsubstrate specificity (6) and shares a common (��)8-barrel fold with SoCel5. This(��)8-TIM-barrel structure is common in 28 GH families, including xylanases, cellulases,mannanases, amylases, and triosephosphate isomerases (CAZy; http://www.cazy.org)(26), and may have evolved through gene duplication events, as in a previously studiedimidazole glycerol phosphate synthase (27). Catalytic amino acids typical of GH5cellulases are also shared between both TeEgl5A and SoCel5 and are in the loops of theprotein surrounding the active site.

Based on this common structure and functionality, we swapped sequence frag-ments between TeEgl5A and SoCel5 to create a chimeric endoglucanase library. Thesesequence swaps occur at the boundary of individual (��) modules. Within eachmodule, there is a �-strand and an �-helix linked together by a ��-loop. For enzymeswith a (��)8 topology, the eight �-strands assemble into a central �-sheet, i.e., thebarrel, which is surrounded by eight �-helices (10). Guided by the enzyme structures,the N-terminal and C-terminal modules of endoglucanases TeEgl5A, SoCel5, and AnCel5

TABLE 1 Schematic structures and molecular masses of the SoCel5-TeEgl5A hybridenzymes

Protein Segment(s)

Positionsa of fragment(s) from:

Mol mass (Da)SoCel5 TeEgl5A

H1 �1�1 47–311 1–54 35,026H2 �2�2 1–49/88–311 58–95 35,445H3 �3�3 1–89/125–311 98–132 34,906H4 �4�4 1–128/160–311 137–167 34,921H5 �1�1��2�2 88–311 1–95 35,605H6 �2�2��3�3 1–49/125–311 58–132 34,921H7 �3�3��4�4 1–89/160–311 98–167 34,801H8 �1�1��2�2��3�3 125–311 1–132 35,485H9 �1�1��2�2��3�3��4�4 160–311 1–167 35,381H10 �5�5��6�6��7�7��8�8 1–159 168–317 34,843aResidue numbering is relative to that of the parent sequence.

Features Imparting Thermal Stability in GH5 Cellulases Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 3

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

were used to design 12 fusion proteins, combinatorially isolating the specific fragmentsthat result in improved thermostability (Table 1; Fig. S1).

Expression and purification of SoCel5 and hybrid enzymes. Recombinant SoCel5was successfully produced in Pichia pastoris GS115 component cells after a 48-hmethanol induction. The enzyme was purified to electrophoretic homogeneity (Fig.S2a) but with an apparent molecular mass higher than the predicted value (35 kDa),with glycosylation likely accounting for the higher molecular mass. After treatment withendo-�-N-acetylglucosaminidase H (endo H) to remove glycosylation, recombinantSoCel5 migrated as a protein band corresponding to its expected molecular mass.

Five (H4, H5, H6, H8, and H9) of the 10 SoCel5-TeEgl5A hybrid enzymes exhibitedendoglucanase activity as measured by the dinitrosalicylic acid (DNS) method. Incomparison to the wild type, these active hybrids showed various protein migrationpatterns in the gel before and after endo H treatment to remove enzyme glycosylation(Fig. S2a). The different numbers of structure-based N-glycosylation sites—four (Asn2,

FIG 1 Sequence alignment of SoCel5, TeEgl5A, and their hybrid enzymes H8 and H9 and three crystallized GH5 cellulases, from Aspergillus niger (PDB 5I78) (11),Thermoascus aurantiacus (PDB 1H1N) (57), and Penicillium verruculosum (PDB 5L9C). Residue numbering below the alignment is according to the sequence fromSoCel5, which is used consistently throughout the text when referring to residue numbers, including for TeEgl5A and the hybrids, to simplify comparison ofinteractions between enzymes. Module elements are labeled above the alignment. Residues highlighted in purple are identical across all sequences, andresidues highlighted in blue are shared in at least 5 enzymes.

Zheng et al. Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 4

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

Asn23, Asn63, and Asn76) for H4, two (Asn21 and Asn44) for H5, two (Asn2 and Asn23)for H6, and one (Asn36) for H8 and H9 —may account for this variation. After the endoH treatment, each hybrid enzyme exhibited a molecular mass equal to its estimatedvalue (Table 1).

Effect of pH. The effect of pH on the activity and stability of wild-type and hybridenzymes was determined using carboxymethyl cellulose-sodium (CMC-Na) as thesubstrate. The wild type and hybrid H4 were optimally active at pH 5.0, while the othershad a pH optimum at 4.0 (Fig. 2a). The enzymes exhibited pH-dependent stability (Fig.2b). The wild type and hybrids H4 to H6 retained long-term activity over a pH range of4.0 to 7.0 (�80% activity) and were largely ineffective and possibly unfolded at pH 8.0to 9.0, but they retained �50% activity at pH 10.0 to 12.0. In contrast, hybrids H8 andH9 had a broader pH stability range, retaining �70% activity at pH 3.0 to 10.0.

Thermal property analysis. To determine the optimal temperatures for SoCel5 andthe hybrids, their enzyme activities at different temperatures (40 to 90°C) were deter-mined after a 10-min reaction with 1% CMC-Na in 100 mM citric acid–Na2HPO4 atoptimal pH. The optimal temperature for wild-type SoCel5 was determined to be 60°C(Fig. 2c). When the (��) module(s) of SoCel5 were replaced with those from TeEgl5A,the temperature optima of all hybrids were lower than the 90°C for TeEgl5A. H4

FIG 2 Enzyme properties of the purified recombinant SoCel5 and TeCel5 and their hybrid enzymes. The relative activities correspondingto 100% with CMC-Na as the substrate are 351 U/mg for SoCel5, 620 U/mg for TeEgl5A, 270 U/mg for H4, 175 U/mg for H5, 101 U/mgfor H6, 721 U/mg for H8, and 917 U/mg for H9. (a) pH-activity profiles tested at the optimal temperature for each enzyme (60°C for SoCel5,50°C for H4, H5, and H6, 80°C for H8, 70°C for H9, and 90°C for TeEgl5A). (b) pH-stability profiles. After incubation of the enzymes at 37°Cfor 1 h in buffers ranging from pH 3.0 to 12.0, the residual activities were determined in 100 mM citric acid–Na2HPO4 buffer at the optimalpH and optimal temperature of each enzyme. (c) Temperature-activity profiles tested at the optimal pH of each enzyme (pH 5.0 for SoCel5and H4 and pH 4.0 for TeEgl5A, H5, H6, H8, and H9). (d) Temperature-stability profiles. Each enzyme was preincubated at 70°C and optimalpH in 100 mM citric acid–Na2HPO4 buffer for different periods of time and subjected to residual activity assay under the optimal conditionsfor each enzyme.

Features Imparting Thermal Stability in GH5 Cellulases Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 5

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

exhibited maximum activity at 50°C, H5 and H6 had a temperature optimum of 60°C,similar to that of the wild type, and H8 and H9 showed optimal activities at 80°C and70°C, respectively. Significant differences were also seen in their degrees of thermo-stability (Fig. 2d). When incubated at 70°C, SoCel5 and hybrids H4 to H6 lost activityrapidly, retaining �20% activity within 10 min. In contrast, H8 and H9 retained �60%activity for 1 h, indicating that in this case, the replacement of the N-terminal (��)module(s) from the mesophilic SoCel5 with those of the thermophilic TeEgl5A improvedthe hybrid’s thermostability.

The thermal stability parameters of the wild-type SoCel5 and SoCel5-TeEgl5A hybridswere also compared (Table 2). The temperature the protein lost 50% of its activity (T50),the melting temperature (Tm), and the half-life (t1/2) of SoCel5 at 55°C were 57°C, 53.6°C,and 0.4 h, respectively. In comparison to the wild type, hybrids H6, H8, and H9 showhigher T50 (7 to 19°C increase) and Tm (7.6 to 22.9°C increase) values and longer t1/2

values (162- to 650-fold) at 55°C, with H8 and H9 being more thermostable.Specific activity and kinetics. GH5 endoglucanases can often act on multiple

substrates, including glucan substrates, as well as birchwood xylan, Avicel, and lami-narin (6, 28, 29). Table 3 shows the activities of the enzyme library on four substrates.Enzymatic digestion was fastest for barley �-glucan, lichenan, and CMC-Na, withcomparatively slow activity when locust bean gum was given as a substrate. Theenzymes demonstrated no activity with birchwood xylan, Avicel, and laminarin, sug-gesting specificity typical of cellulolytic endoglucanases. The specific activities of SoCel5against barley �-glucan, lichenan, and CMC-Na were higher than those of hybrids H4 toH6, which had replacements �4�4 (H4), �1�1��2�2 (H5), and �2�2��3�3 (H6), butmuch lower than those of hybrids H8 and H9 and TeEgl5A. These results indicate thatintroduction of the N-terminal three (H8) and four (H9) blocks of (��) modules ofTeEgl5A improved the catalytic performance of SoCel5. H9 exhibited the highestspecific activities among all enzymes against barley �-glucan, lichenan, and CMC-Na,which were 20%, 116%, and 160% higher, respectively, than those of the parentalSoCel5 enzyme. The improved catalysis is in some cases nonadditive, as both H8 and H9demonstrated improved performance on CMC-Na relative to the performance of bothparent enzymes.

To better understand the effects of different motifs from TeEgl5A on the catalyticperformance of SoCel5, kinetic studies of the wild-type SoCel5 and SoCel5-TeEgl5A

TABLE 2 Thermodynamic properties of SoCel5, TeEgl5A, and their hybrid enzymes

Enzyme T50 (°C) Tm (°C) t1/2 (h)a

SoCel5 57 � 1 53.6 � 1.2 0.40 � 0.05TeEgl5A 87 � 2 78.1 � 1.5 76 � 0.5H4 58 � 2 51.2 � 1.0 0.06 � 0.01H5 56 � 1 49.3 � 1.1 0.50 � 0.04H6 64 � 1 61.2 � 1.0 65 � 2.5H8 76 � 2 76.5 � 1.3 260 � 3.5H9 72 � 2 70.1 � 1.2 96 � 1.5at1/2 was determined at 55°C.

TABLE 3 Specific activities of SoCel5, TeEgl5A, and their hybrid enzymes on differentsubstrates

Enzyme

Sp act (U/mg) on:

Barley �-glucan Lichenan CMC-Na Locust bean gum

SoCel5 1,540 � 10 1,440 � 80 350 � 6 82 � 6TeEgl5A 2,690 � 20 3,290 � 30 600 � 30 96 � 13H4 1,130 � 80 1,420 � 70 270 � 10 71 � 1H5 550 � 20 630 � 30 175 � 2 29 � 2H6 103 � 2 160 � 2 101 � 1 1.4 � 0.1H8 1,610 � 50 2,550 � 50 720 � 30 55 � 3H9 1,840 � 90 3,100 � 100 920 � 50 74 � 6

Zheng et al. Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 6

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

hybrids were performed using CMC-Na as the substrate under the optimal reactionconditions for each enzyme (Table 4). The Km and Vmax values of SoCel5 were4.9 � 0.3 mg/ml (mean � standard deviation) and 647 � 46.7 mol/min/mg, respec-tively, and its catalytic efficiency was kcat/Km 76.2 � 1.8 ml/s/mg, which is lower thanthat of the endoglucanase (118 ml/s/mg) from Penicillium purpurogenum (30). Thekinetic values of the hybrids had the same trends as the specific activities. In compar-ison to the SoCel5 parent, hybrids H4 to H6 showed decreased substrate affinities(higher Km values) and reduced reaction velocities and turnover rates (lower Vmax andkcat values), while hybrids H8 and H9 showed greater substrate affinities and reactionvelocity and turnover rates. The catalytic efficiencies of hybrids H4 to H6 decreased to17.4 to 51.4% of that of SoCel5, and those of H8 and H9 increased up to 276%. The H8and H9 enzymes specifically are in some sense even more efficient than the thermo-philic TeEgl5A parent, as they combine the high affinity of SoCel5 with the higherturnover rate of TeEgl5A for a 2- to 3-fold improvement in catalytic efficiency (kcat/Km)relative to that of either parental enzyme. These findings indicate that the replacementof different (��) modules from TeEgl5A could either enhance or prove deleterious tothe function of SoCel5, depending on the specific replacements made. The primarydifferences in the five SoCel5-TeEgl5A hybrids in catalytic performance (substratebinding, enzyme catalysis, and product dissociation) may be attributed to their struc-tural differences. Hybrids H4 to H6 had only one or two (��) module(s) from TeEgl5A,while hybrids H8 and H9 contained three or four (��) exogenous modules (Table 1).Hybrids H5 and H8 had one module difference (�1�1��2�2 versus �1�1��2�2��3�3)but differed in temperature optimum (60°C versus 80°C), degree of thermal stability,and level of catalytic performance. Based on these results alone, we conjecture that the�3�3 module may play a key role in protein structure and catalysis. However, hybrid H3(containing the �3�3 of TeEgl5A alone) did not show improvements in the propertiesexamined in this study. Moreover, hybrid H4, containing partial segments of thereplaced modules of H8 and H9, had no improvement in stability or catalysis.

Functional verification by AnCel5 and its hybrids. To verify the common effect ofthe (��)1–3 and (��)1– 4 modules of TeEgl5A, the corresponding part of another GH5endoglucanase from A. nidulans was also replaced (11). Two hybrid enzymes(AnCel5-H1 and AnCel5-H2) were constructed and produced in P. pastoris GS115, butonly AnCel5-H2 showed endoglucanase activity. After purification to homogeneity (Fig.S2b), wild-type AnCel5 and the hybrid AnCel5-H2 were biochemically characterized.Both AnCel5 and AnCel5-H2 were optimally active at pH 4.0 (Fig. S3a). However,AnCel5-H2 showed adaptability and stability over higher temperatures. It had anoptimal temperature of 80°C, 10°C higher than that of wild-type AnCel5 (Fig. S3b).AnCel5 was stable at temperatures of �60°C, while the hybrid AnCel5-H2 retainedstability at 70°C (Fig. S3c). Moreover, AnCel5 lost activity rapidly at 70°C and 80°C(retaining �10% within 20 min), while under the same conditions, AnCel5-H2 retained�30% activity after a 1-h incubation (Fig. S3d). The T50, Tm, and t1/2 (70°C) values ofAnCel5 and AnCel5-H2 were 70 � 2°C and 78 � 1°C, 83.5 � 1.0°C and 87.0 � 1.0°C, and0.6 � 0.1 h and 9.0 � 0.6 h, respectively. When using CMC-Na as the substrate, AnCel5

TABLE 4 Kinetic parameters of SoCel5, TeEgl5A, and their hybrid enzymes with CMC-Na asthe substrate

Enzyme Km (mg/ml) Vmax (�mol/min/mg) kcat (turnover, s�1) kcat/Km (ml/s/mg)

SoCel5 4.9 � 0.3 650 � 50 380 � 30 77 � 2TeEgl5A 7.4 � 0.6 1,010 � 30 600 � 10 81 � 5H4 7.3 � 0.1 352 � 7 204.8 � 4 28.1 � 0.3H5 5.1 � 0.3 251 � 9 149 � 5 29.2 � 0.9H6 7.3 � 0.4 167 � 13 100 � 10 13.4 � 0.2H8 4.2 � 0.3 1,200 � 40 710 � 30 170 � 7H9 4.0 � 0.2 1,450 � 110 850 � 60 210 � 40AnCel5 4.2 � 0.2 2,400 � 40 1,400 � 20 330 � 30AnCel5-H2 3.4 � 0.2 2,490 � 30 1,460 � 20 430 � 10

Features Imparting Thermal Stability in GH5 Cellulases Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 7

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

and AnCel5-H2 showed specific activities of 1,818� 26 U/mg and 2,800� 34 U/mg,respectively, and similar Vmax (2,397� 57 �mol/min/mg versus 2,486� 36 �mol/min/mg) and kcat (1,402� 35/s versus 1,462� 47/s) values but different Km values(4.2 � 0.3 mg/ml versus 3.4 � 0.2 mg/ml). When combined with the previous results,these results show that selective recombination of different related enzymes canindeed be an effective strategy for improving the thermal stability and catalyticefficiency of GH5 fungal cellulases.

Mechanism of improved thermostability. The evidence presented indicates thatthere are structural elements unique to the N-terminal region of the thermophilicTeEgl5A that enhance the performance of the enzyme at high temperature. Based onthe melting temperatures determined (Table 2), a region of particular interest would bethe interface between �2�2 and �3�3, which when replaced by the TeEgl5A equivalentas in H6, raises the melting temperature relative to that of unaltered SoCel5 by nearly10°C. Additional 10 to 15°C increases in the melting temperature when �1�1 fromTeEgl5A is also included, as in H8 and H9, but not when it is independently replaced,as in H1 and H5, suggest that the �1�1 module is also involved in the denaturationprocess of these hybrid enzymes.

To explore which specific interactions may participate in the increased meltingpoints of specific chimeras, conventional equilibrium MD simulations were performedfor models of H8 and H9, as well as both parent enzymes. By varying the simulationtemperature from 25°C (298 K) to near the melting point of 70°C (343 K) and beyondto 125°C (398 K), we tracked how these interactions changed with temperature over thecourse of the five replicates for each combination of temperature and protein. Bycorrelating specific interactions with temperature changes across the conformationalensemble created by simulation, we resolved the improved hydrophobic packing at theinterface between modules 2 and 3. Since single simulations were only 200 ns long,only minimal thermal denaturation was observed. Together, the results from the MDsimulations provide mechanistic insight as to the denaturation process and how it isarrested in part in the H8 and H9 hybrids.

Barrel stabilization by hydrogen bond networks. A commonly invoked mecha-nism to improve protein thermostability is rigidifying mutations (31), which result insignificantly lower fluctuations for specific residues. Using MD simulations, we coulddirectly test for reduced fluctuations at specific residue positions by computing the rootmean squared fluctuation (RMSF), which measures the mean fluctuation away from theaverage position of that residue (Fig. 3). We found that there was a consistent reductionin fluctuation at elevated temperature only for the original thermophilic TeEgl5A,particularly in structured regions within the �-sheets. The hybrid enzymes, in contrast,had fluctuations in line with what was observed in SoCel5, even in regions where thesequence was identical to that of TeEgl5A (�1�1-�3�3 for H8 and �1�1-�4�4 for H9).Thus, the increased thermostability of H8 and H9 is not strictly due to rigidifyingmutations.

Instead, the reduced RMSFs for the �-sheets within TeEgl5A imply that there arespecific interactions formed within the barrel of TeEgl5A that are not present in SoCel5or its hybrids. From the sequences alone, it is not clear which residues are in closeproximity and might interact across the barrel. The TeEgl5A sequence introduces anumber of charged residues within the barrel relative to the sequence of SoCel5, someof which were determined by pKa estimation tools (32) to be protonated at physio-logical pH, based on their inaccessibility to water and possible hydrogen bonds thatcould be made with neighboring residues. The network of interactions that is formedin TeEgl5A (Fig. 4B) directly connects more structural elements relative to the interac-tions seen in SoCel5 (Fig. 4A), reducing the increases in fluctuation within the centralbarrel as the temperature rises (Fig. 3). In the hybrid enzymes, only some of theseinteractions are retained. There is effectively a mutation of Q to D at position 94 (Q94D)in both H8 and H9 due to the sequence they inherit from TeEgl5A. This change relativeto the sequence of SoCel5 is sufficient to provide R52 a strong binding partner and

Zheng et al. Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 8

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

significantly lower the fluctuations of the R52 residue in the hybrids, as indicated by theblack arrows in Fig. 3.

Other interactions found in TeEgl5A that further stabilize the central barrel aremissing from the hybrids. One example occurs at position 201, where H8 and H9 retainQ201 instead of having the E201 found in TeEgl5A. As a result, neighboring chargedresidues do not form extended hydrogen bonding networks, as they do in TeEgl5A (Fig.4B), and instead show a lack of interaction similar to that in SoCel5 (Fig. 4A). At lowtemperatures, the weaker hydrogen bond networks formed with N13 and sporadicinteractions between Q94 and Q201 are sufficient to stabilize the barrel. However,without the central barrel stabilization brought about by the additional hydrogenbonds seen in TeEgl5A, the barrel exhibits higher fluctuations with increasing temper-ature. The extensive TeEgl5A interaction network increases the thermostability of thebarrel complex and may be an additional avenue by which thermostability could befurther improved, similar to efforts in other fungal cellulases to add hydrogen bonds toimprove thermostability (13, 33).

Specific interactions in �2�2 and �3�3. To narrow down what subdomain com-ponents are interacting, we first evaluated the hydrogen bonds within the first 4module sets (Table 5). In this analysis, we see that SoCel5 actually creates hydrogenbonds more frequently within the first 4 modules than do the hybrids or the thermo-philic enzyme. This difference shrinks when the same analysis is conducted on thehigher-temperature simulations (Tables S1 and S2), indicating that these interactionsare not as stable overall. In part, this may be due to the unexpected hydrogen bondformed between Y77 and V53 in TeEgl5A and hybrids H8 and H9 (Fig. 4D) that is notpresent in SoCel5 (Fig. 4C). This interaction is the only direct hydrogen bond formedthat goes between a helical segment and a �-strand within the first four modules of the

FIG 3 Comparative per-residue root mean square fluctuations (RMSFs) across all trajectories. The RMSFfor each simulated GH5 is presented on three graphs (black, SoCel5; red, H8; blue, H9; gray, TeEgl5A), onefor each temperature, as indicated. To highlight the disparity of RMSFs between structural elements andthe intervening loops, the eight beta-strand regions are highlighted in pink and the eight alpha-helicalregions are highlighted in green, as well as labeled in the 398 K graph. The elevated RMSF for SoCel5 atresidue 52 is indicated by black arrows.

Features Imparting Thermal Stability in GH5 Cellulases Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 9

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

enzyme (Table 5). When combined with the adjacent rigidifying interactions in thebarrel core surrounding R52, this provides part of a mechanism by which sequencereplacement improves thermostability.

The hydrogen bond from Y77 to V53 is made possible by other aromatic residues inthe vicinity subtly perturbing the relative orientation of the �2 and �3 helices. In SoCel5,residue 123 is a tyrosine, which interacts with T78 to satisfy its hydrogen bondingrequirements in an otherwise hydrophobic region of the protein (Fig. 4C). The effectiveY123F mutation in the hybrids and TeEgl5A, coupled to N122 hydrogen bonding to S78,creates a hydrophobic pocket full of favorable �-stacking interactions (Fig. 4D). Thestacking interactions maintain a favorable environment for Y77 without causing a helixrotation that would destroy the structure, in turn causing the carbonyl of V53 to havean alternate hydrogen bonding partner. These interactions together mean that thehydrophobic packing at this interface improves relative to that of the mesophile, aphenomena that has been noted in interaction clusters in other cellulases (34).

An alternative method of quantifying these hydrophobic interactions is to deter-mine the number of contacts between different structural elements within the enzyme.The overall contact structure highlights the similarity of the folds in all of the enzymesconsidered here (Fig. S4). However, the most revealing aspect of contact analysis occurswhen comparing the numbers of contacts across different temperatures (Fig. 5). Asexpected, the number of observed contacts tends to decrease at high temperature dueto the increased fluctuation computed previously (Fig. 3). The singular clear exceptionis the contacts between �2 and �3, which actually increase with temperature in the

FIG 4 Simulation snapshots. In the barrel assemblies of SoCel5 (A) and TeEgl5A (B), we highlight specificinteractions around the R52 residue that is present for each enzyme. For visual clarity, only heavy atomsand polar hydrogens for the selected residues that form an extended hydrogen bond network with R52are shown. Along the interface between �2�2�3�3, sequence differences between SoCel5 (C) and TeEgl5A(D) lead to different hydrogen bonds and hydrophobic packing arrangements. The hydrogen bondinteractions are shown as green dashed lines.

Zheng et al. Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 10

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

hybrids and TeEgl5A. The hydrophobic effect strengthens at higher temperature (35),driving the observed aromatic packing within the simulations.

DISCUSSION

Protein structure and function evolve through sequence changes, substitutions,duplications, insertions, and deletions, including rearrangement or recombination ofshort fragments as we did here, building on earlier work showing that folded hybriddomains can be generated by shuffling polypeptide segments (36). Based on priorexperimental and structural studies, which include specific research in glycoside hy-drolases (37, 38) and whose results suggest that the common (��)8 barrel or TIM barrelevolved from an ancestral half-barrel through a series of duplication, fusion, anddiversification events (20, 39), our work here is in effect accelerated evolution with aparticular design goal in mind. The research here is designed to combine the highsubstrate affinity of a mesophilic enzyme with the thermostability of a thermophilicenzyme to produce more efficient enzymes for specific substrates.

The key to the success of the improved enzyme variants was replacing theN-terminal half-barrel, as the C-terminal half-barrel hybrid was not functional. This hasprecedents elsewhere in the literature for TIM barrel enzymes. In a previous study,Sharma and coworkers reported a stable and active chimera, CelB-CelCCA, whichshowed maximum activity at 70°C and was created by fusing the N-terminal half-barrelof the thermophilic CelB (maximum activity at 95°C) and C-terminal half-barrel ofmesophilic CelCCA (maximum activity at 50°C) (37). Furthermore, Numata et al. con-structed chimeric isopropylmalate dehydrogenases by connecting fragments from a

TABLE 5 Hydrogen bond propensities at 298 K across all simulations for residues within the �1– 4 or�1– 4 modules, excluding hydrogen bonds involved in helical interactionsa

aA hydrogen bond is counted only if the heavy atoms are within 3.2 Å and the hydrogen is no more than 30 degreesremoved from the direct line between the heavy atoms. To help quickly identify the structural element in which eachresidue is found, the residue identifiers are color coded. Module 1 uses black, module 2 uses blue, module 3 uses green,and module 4 uses red. The shade of the color indicates whether it is in a � element (dark) or an � element (light).

Features Imparting Thermal Stability in GH5 Cellulases Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 11

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

thermophilic and a mesophilic parental enzyme (40). They found that the thermalstability of the chimeric enzymes was nearly proportional to the fraction of thesequence coming from the thermophilic enzyme, suggesting that amino acid residuescontributing the thermal stability distribute themselves in the N-terminal half. Togetherwith our results, this suggests that the N-terminal half-barrel determines the thermo-stability of TIM barrel proteins, possibly by being the first part to unfold completely atthe melting transition, although this was not observed over the short simulationtimescales.

The observed high activity of two hybrid enzymes on specific substrates was alsonot guaranteed. Recombination of the segments between related enzymes oftenresults in hybrids with diminished activity. For example, Hosseini-Mazinani and cowork-ers created 18 hybrid genes by replacing the coding region of the P. vulgaris�-lactamase gene with the equivalent portions from the RTEM-1 gene (41). Most ofthese hybrids produced inactive proteins, and a few hybrid enzymes had partial or traceactivity. Even though the previously mentioned chimera CelB-CelCCA displayshyperthermophile-like structural stability, the chimera’s activity is significant lower thanthose of the parental enzymes, CelB and CelCCA (37). Similarly, in our research, half ofthe 10 hybrids showed no activity even though they were constructed with a meth-odology similar to that used for the other 5 that were active, implying that moreresearch will need to be done to elucidate the determinants of chimeric proteinfunction.

To our surprise, two hybrid enzymes (H8 and H9) demonstrated increases in theirspecific activity and catalytic efficiency using carboxymethylcellulose sodium (CMC-Na)as a substrate. Normally, mutants with increased stability often lose catalytic efficiencybecause flexibility is required for enzyme activity, whereas structural rigidity improvesthermostability (31, 42). In contrast to this, both H8 and H9 showed improved thermo-stability and catalytic efficiency, which reduces the enzyme loading (and thereby thecost) required for production. We suspect that the systematic nature of the SCHEMAreplacement and high structural homology of the TIM barrel proteins resulted incorrectly positioned amino acids that were in the appropriate position for substratebinding and catalytic bond cleavage after functional domain recombination.

Parental enzymes that are sufficiently related lend themselves to the construction of

FIG 5 Atomic contact changes between different structural elements in going from 298 K to 398 K. Eachof the four enzymes studied is labeled in the top right of each panel, with the color scale defined on the farright. Contacts were defined between individual atom pairs that were within 5 angstroms during simulationand weighted according to distance. The weighting function used was C(t) � �pairs[1/1 � e5(d � 4 Å)], whered is the distance between individual atoms in the pair.

Zheng et al. Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 12

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

hybrid proteins. Recombination segments or subdomains of proteins are highly similar,and thus, the interaction required for the proper structure and function of hybrids willbe retained. Then, recombinants are analyzed in an attempt to identify determinantsresponsible for parameters such as thermostability or activity (43). Based on the identityof the modules replaced in the hybrids exhibiting enhanced thermostability, H8, H9,and to some extent, H6, the reasonable conclusion is that the interfaces between the�2, �2, �3, and �3 elements within SoCel5 are stabilized through hybridization withTeEgl5A. In addition, with the results showing the mechanism behind the melting pointimprovement of the H8 and H9 hybrids relative to that of the progenitor SoCel5, we canbegin to speculate on the thermal denaturation mechanism of SoCel5. Since the hybridenzymes strengthened hydrophobic interactions between the second and third mod-ule to improve the melting point, this suggests that native SoCel5 unfolding is initiatedby water disrupting the packing between �2 and �3. One potential mechanism seen insimulation is T78 rotating away from its hydrogen bonding interaction with Y123 whenit has the thermal energy to do so, drawing water to interact closely with Y123. In theH6, H8, and H9 hybrids, increasing the temperature strengthens the aromatic hydro-phobic interactions at this same interface, raising the melting point until anotherstructural element denatures and disrupts the structure.

The order of the relative melting points between H6, H8, and H9 can be used toinform the order of denaturation. Hybrid H6 has a lower melting point than H8 or H9while sharing the thermophilic version of modules 2 and 3, but not that of module 1.This suggests that a structural element in module 1 unfolds first, likely �1 due to itsgreater accessibility to solution. After �1 unfolds, �2 would then not be packed againstit and might then drift away from �3, perturbing the packing between �2 and �3 andleading to enzyme unfolding. Our analysis finds no specific interactions with theremainder of the enzyme that would stabilize �1. Instead, our hypothesis is muchsimpler and is motivated by sequence changes at the N-terminal end of �1. Residue 35of TeEgl5A is a proline, compared with a tyrosine in SoCel5. Prolines at the N-terminalend are known to raise the melting point of �-helices (44–46). Similarly, residue 36 ofTeEgl5A introduces an additional N-glycosylation site relative to the sequence ofSoCel5. Glycosylation is also known to increase the thermostability of �-helices (47, 48).Together, these changes maintain the structure of �1 at a higher temperature, which inturn protects the packing of the �2 and �3 helices.

This does not, however, mean that �1 is the first helix to unfold. Even consideringthe short simulated timescales, we observed a loss of structure in �5 even at modesttemperature, as evidenced by the high RMSF in that part of the protein (Fig. 3). It maybe that this helix is natively in equilibrium between its folded and unfolded states andthat the adjacent residues were uniquely chosen to form compensatory interactionswith the unfolded helix, preserving the overall structure. Such compensatory interac-tions would provide a mechanism for the melting point differential between H8 and H9,since only in H9 is the native �4-�5 interface disrupted by the sequence changes made.However, another possible explanation is that the unfolding of �5 is a modeling artifactof adding in two additional alanines relative to the sequences of the homologouscrystallographic models.

In summary, through combinatorial swapping of sequence elements between amesophilic and a thermophilic endoglucanase, we created two enzymes with higherefficiency than either of the parental enzymes. These successful hybrid enzymesdemonstrate increased thermostability relative to that of the mesophilic parent whilestill being less thermostable than a true thermophile. The high efficiency and improvedthermostability may be useful for reducing enzyme loading within industrial processes.Through the companion MD simulations performed on a ladder of temperatures, theinteractions that lead to enhanced thermostability have been identified on the �2-�3

interface, which improves residue packing. Recapitulating the R52 interactions withsurrounding anionic residues identified in the thermophilic TeEgl5A may be a furthermethod of stabilizing the newly created hybrid enzymes.

Features Imparting Thermal Stability in GH5 Cellulases Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 13

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

MATERIALS AND METHODSStrains, plasmids, and chemicals. The donor S. opalus strain was cultivated at 28°C in an inducing

medium containing 5 g/liter (NH4)2SO4, 1 g/liter KH2PO4, 0.5 g/liter MgSO4 · 7H2O, 0.2 g/liter CaCl2,10 mg/liter FeSO4 · 7H2O, 30 g/liter corncob, 30 g/liter soybean meal, and 30 g/liter wheat bran. Esche-richia coli Trans I-T1 (TransGen, Beijing, China) was used for gene cloning and construction of the hybridenzymes. The vector pPIC9 and Pichia pastoris GS115 from Invitrogen (Carlsbad, CA) were used forenzyme expression. Yeast extract-peptone-dextrose (YPD), minimal dextrose, buffered glycerol complex(BMGY), and buffered methanol complex (BMMY) were prepared according to the Pichia expression kitmanual (Invitrogen). FastPfu DNA polymerase was purchased from TransGen, restriction endonucleasesfrom Fermentas (Burlington, ON, Canada), T4 DNA ligase from New England Biolabs (Hitchin, UK), andBglII from TaKaRa (Kyoto, Japan). The substrates carboxymethyl cellulose-sodium (CMC-Na), barley�-glucan, Avicel, laminarin, birchwood xylan, and locust bean gum were supplied by Sigma-Aldrich (St.Louis, MO). Lichenan was purchased from Megazyme (Wicklow, Ireland).

Cloning and sequence analysis of the gene SoCel5. The genomic DNA of S. opalus was extractedusing a DNA isolation kit (Tiangen, Beijing, China). The total RNA was extracted, reverse transcribed, andused as a template for cDNA amplification as described by Zhao et al. (49). Using the genomic DNA ofS. opalus as a template, the core region of the endoglucanase-encoding gene SoCel5 was amplified witha degenerate primer set, GH5-F/GH5-R (Table 6), and its 5=- and 3=-flanking regions were obtained by

TABLE 6 Primers used in this study

Primer Sequence (5=¡3=)a

GH5-F GAYCCNCAYAAYTWYGGGH5-R TCNARRTAYTGRTGCATYTF (Socel5-F) CCGAATTCCAAAACTCGACCAAGAAGCTGAAATGGCR (Socel5-R) TAGCGGCCGCTTACACAAAATACGACTTAATCTTCGGCAAAATCCus-1 CCACGAGTTCCCCTCAGGGGTGus-2 CGTTGTTTGTGTCGAAGACGACGAGGTCds-1 CGACCTCGTCGTCTTCGACACAAACAACds-2 GAACTCGTGGACGGGCGCTTGGA (Teegl5A-F) CCGAATTCGCTCCAGTCAAGGAGAAGGGCATCAAGB (Teegl5A-R) TAGCGGCCGCTCATTCGAGGTACGGCTTCAGCAAF1 CCAGATCCTCATCGACCAGGGCGTCAACCACTTCR1 GAAGTGGTTGACGCCCTGGTCGATGAGGATCTGGF2 GGCCCAAGGCGTCAACATCTTCCGCGTGCCGTR2 ACGGCACGCGGAAGATGTTGACGCCTTGGGCCF3 CTACATTACCAGCCATATAACCGCAAAAGGCGR3 CGCCTTTTGCGGTTATATGGCTGGTAATGTAGF4 TAACCGCAAAAGGCGCGTCGGCAGTGATTGACCCGR4 CGGGTCAATCACTGCCGACGCGCCTTTTGCGGTTAF5 TATTGCGTCCAACTTTGCGGACAACGACCTCGTCGTCR5 GACGACGAGGTCGTTGTCCGCAAAGTTGGACGCAATAF6 CAAAGACAACGACCTCGTCATTTTCGACACGAACAR6 TGTTCGTGTCGAAAATGACGAGGTCGTTGTCTTTGF7 CGGCATCCGCGCCGCGGGCGCAACGAAGCAATACR7 GTATTGCTTCGTTGCGCCCGCGGCGCGGATGCCGF8 CTACATTACCAGCCATATAACCGCAAAAGGCGR8 CGCCTTTTGCGGTTATATGGCTGGTAATGTAGF9 GGCCCAAGGCGTCAACATCTTCCGCGTGCCGTR9 ACGGCACGCGGAAGATGTTGACGCCTTGGGCCF10 TATTGCGTCCAACTTTGCGGACAACGACCTCGTCGTCR10 GACGACGAGGTCGTTGTCCGCAAAGTTGGACGCAATAF11 TAACCGCAAAAGGCGCGTCGGCAGTGATTGACCCGR11 CGGGTCAATCACTGCCGACGCGCCTTTTGCGGTTAF12 CGGCATCCGCGCCGCGGGCGCAACGAAGCAATACR12 GTATTGCTTCGTTGCGCCCGCGGCGCGGATGCCGF13 TATTGCGTCCAACTTTGCGGACAACGACCTCGTCGTCR13 GACGACGAGGTCGTTGTCCGCAAAGTTGGACGCAATAF14 GACAATGTCATTTTCGACACGAACAACGAATACCACR14 GTGGTATTCGTTGTTCGTGTCGAAAATGACATTGTCF15 CGGCATCCGCGCCGCGGGCGCAACGAAGCAATACATCR15 GATGTATTGCTTCGTTGCGCCCGCGGCGCGGATGCCGC (Ancel5-F) TCGAATTCGTGCCTCATGGTCCTGGACATAAGD (Ancel5-R) TAGCGGCCGCTCAAAGATATGCCTCCAGGATATCCAGCF16 CACTATTGCGTCCAACGATAACGACCTGGTCATGTTTGACACTAAR16 ATGACCAGGTCGTTATCGTTGGACGCAATAGTGTGCCAGAAAGF17 ACGGCATCCGCGCCGCGGGTGCGACCAGCCAGTACATCTTCR17 TACTGGCTGGTCGCACCCGCGGCGCGGATGCCGTCGATaRestriction sites are underlined.

Zheng et al. Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 14

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

thermal asymmetric interlaced PCR (TAIL-PCR) with four arbitrary degenerate primers from the TaKaRagenome-walking kit and four nested specific primers (us-1, us-2, ds-1, and ds-2) (Table 6) designed basedon the core region sequence of SoCel5 (50). The PCR products were ligated with pEASY-T3 vector forsequencing and assembled to give the full-length SoCel5 (GenBank accession number ARO48344). Theexpression primers Socel5-F/Socel5-R with EcoRI and NotI restriction sites (Table 6) were used to amplifythe cDNA fragment coding for mature SoCel5 without the putative signal peptide sequence. The PCRproduct was purified using a gel extraction kit (Omega Bio-tek, Norcross, GA) and then ligated into thepPIC9 expression vector to produce the recombinant expression vector pPIC9-Socel5.

The DNA and amino acid sequences were analyzed using the BLASTx and BLASTp programs(https://www.ncbi.nlm.nih.gov/BLAST/) (51), respectively. The introns, exons, and transcription initiationsites were predicted using the GENSCAN Web Server (http://genes.mit.edu/GENSCAN.html) (52). SignalP3.0 was used to predict the signal peptide sequence (http://www.cbs.dtu.dk/services/SignalP/) (53).Sequence assembly and estimation of the molecular mass and pI of the mature peptide were performedusing Vector NTI Suite 10.0 software (Invitrogen).

Production and purification of the recombinant SoCel5. The recombinant plasmid pPIC9-Socel5was linearized by BglII and introduced into P. pastoris GS115 competent cells by electroporation.Transformants were selected on plates of minimal dextrose medium at 30°C for 2 days. The positiveclones were grown in shaker tubes containing 3 ml of BMGY at 30°C for 48 h, followed by cell collectionand enzyme induction in 1.5 ml BMMY containing 0.5% methanol at 30°C for 72 h. The culture super-natant of each transformant was collected by centrifugation at 12,000 g for 10 min at 4°C andexamined by both SDS-PAGE and an endoglucanase activity assay. The positive transformants were alsoverified by colony PCR and sequencing. For scaled-up cultivation, the transformant with the highestendoglucanase activity was grown in 1-liter Erlenmeyer flasks containing 400 ml of BMGY at 30°C for 48 hwith an agitation rate of 200 rpm. Cells were harvested and resuspended in 200 ml of BMMY containing0.5% (vol/vol) methanol for a 48-h induction at 30°C.

The culture supernatants were collected by centrifugation at 12,000 g for 10 min at 4°C, followedby ultrafiltration using a Vivaflow 50 ultrafiltration membrane with a molecular weight cutoff of 10 kDa(Vivascience, Hannover, Germany). The crude enzyme was applied to a HiTrap Q HP anion exchangecolumn (Amersham Biosciences, Uppsala, Sweden) equilibrated with a 10 mM phosphate buffer of pH7.5. A linear NaCl gradient of 0 to 1 M was used to elute the proteins. The apparent molecular mass andpurity of purified recombinant SoCel5 were estimated by SDS-PAGE. Endo-�-N-acetylglucosaminidase H(endo H) from New England Biolabs was used to remove N-glycosylation according to the manufacturer’sinstructions.

Enzyme characterization. The endoglucanase activity was determined by using the dinitrosalicylicacid (DNS) method (54). Reaction mixtures containing 100 �l of properly diluted protein solution and900 �l 1% (wt/vol) CMC-Na were incubated at pH 5.0 (100 mM citric acid–Na2HPO4) and 60°C for 10 min,followed by the addition of 1.5 ml DNS solution and 5 min in a boiling water bath. When the reactionmixtures cooled to room temperature, the absorbance at 540 nm was measured. The standard curve forcalibrating the enzyme activity was determined using 0.25 to 3.5 �mol glucose. One unit of enzymeactivity was defined as the amount of enzyme required to release 1 �mol reducing sugars per min.Specific activity was defined as enzymatic units per milligram protein.

The optimal pH for SoCel5 was determined at 60°C in 100 mM citric acid–Na2HPO4 buffer with pHranges of 3.0 to 8.0. The optimal temperature was determined at pH 5.0 at a temperature range of 50to 90°C. The pH stability was determined by measuring the residual activity at pH 5.0 (100 mM citricacid–Na2HPO4) and 60°C for 10 min after 1 h of incubation at pH 3.0 to 12.0 and 37°C without CMC-Na.Thermostability was investigated after incubation of the samples at pH 5.0 and 70°C for different periodsof time. Residual activity was measured as described above.

Substrate specificity of SoCel5 was determined by using 1% CMC-Na, barley �-glucan, lichenan,birchwood xylan, Avicel, laminarin, or 0.5% locust bean gum as the substrate. The kinetic parameters Km,kcat, Vmax, and kcat/Km were estimated from the SoCel5 activities at pH 5.0 and 60°C for 5 min by using 1to 10 mg/ml CMC-Na as the substrate. GraphPad Prism 6.0 was used to calculate the values by using theLineweaver-Burk plot.

Design and construction of the hybrid enzymes. SoCel5 shared 51% and 47% sequence identitywith the thermophilic endoglucanase TeEgl5A (GenBank accession number KF680302) and its N-terminal(��)4 modules (Fig. 1), respectively. By using the fusion protein method, the N-terminal (��)1– 4 module(s)or the C-terminal (��)5– 8 modules of TeEgl5A were introduced into the corresponding parts of SoCel5(Table 1), and a total of 10 hybrid enzymes were constructed. To further verify the functional roles of theN-terminal sequence of TeEgl5A in another GH5 endoglucanase, AnCel5 (GenBank accession numberAAG50051) from Aspergillus niger (11), sharing 56% identity with SoCel5 and 67% identity with TeEgl5A,was selected and hybrid enzymes were constructed by replacing the N-terminal (��)1–3 and (��)1– 4

modules, analogous to hybrids H8 and H9 with SoCel5.All the fusion proteins were obtained by a two-step overlap-extension PCR (Fig. S1 in the supple-

mental material). In the first-step PCR, parallel reactions were performed to amplify the objective DNAfragments using the recombinant plasmids pPIC9-Teegl5A, pPIC9-Socel5, and pPIC9-Ancel5 as templatesand primers. The specific recombinations are listed in Table 6. The second-step PCR was used to amplifythe final DNA products with the first-step PCR products as templates and F/A and R/B/D primer sets(Table 6). The 50-�l PCR mixture contained 1 �l of each primer, 1 �l of Fastpfu Fly DNA polymerase(TransStart, Beijing, China), 5 �l of deoxynucleoside triphosphates (dNTPs), 10 �l of Fastpfu fly buffer, 3 �lof MgSO4, 1 �l of template DNA, and 28 �l of double-distilled water (ddH2O). The PCR protocolcomprised an initial denaturation at 95°C for 5 min, followed by 30 cycles of 94°C for 30 s, annealing at

Features Imparting Thermal Stability in GH5 Cellulases Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 15

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

60°C for 30 s, and elongation at 72°C for 60 s, with a final extension at 72°C for 10 min. The PCR productswere purified, digested with EcoRI and NotI, ligated into the pPIC9 expression vector, and sequenced. Theproduction, purification, and characterization of hybrid enzymes followed the same procedures used forrecombinant SoCel5.

Thermal stability analysis. For short-term thermostability analysis, the purified recombinant wild-type and hybrid enzymes were incubated at 70°C and/or 80°C and optimal pH for 1 h without substrate.Samples were taken at specific time points for the endoglucanase activity assay, which was performedunder optimal conditions for each enzyme. Three thermodynamic parameters, T50, t1/2, and Tm, were usedto compare the thermal properties of SoCel5, AnCel5, and their hybrid enzymes. T50 was defined as thetemperature at which a 30-min incubation caused the protein (0.1 mg/ml) to lose 50% activity, while t1/2

was defined as the half-life of an enzyme at 55°C (for SoCel5 and its hybrid enzymes) or 70°C (for AnCel5and its hybrid enzymes). Due to cooperative unfolding observed at long timescales for some chimericenzymes, the reported t1/2 is based on bracketing the 50% activity point with activity measurement timepoints. An example of the differences in t1/2 that arise from the cooperative unfolding relative to atraditional exponential fit is given in Fig. S5. Differential scanning calorimetry (DSC) was used todetermine the Tm values. The Nano DSC (TA Instruments) was run at a heating-and-scanning rate of1°C/min over a temperature range of 30 to 90°C. Each sample contained 0.25 mg/ml protein in 10 mMcitric acid–Na2HPO4 buffer (pH 7.5). The test was repeated at least twice.

MD simulation. To determine the specific molecular interactions underlying the observed improve-ment in thermostability, a series of equilibrium classical molecular dynamics (MD) simulations ofmodelled GH5 structures were performed using NAMD 2.12 (55). Four GH5 models were constructed, oneeach for the SoCel5 and TeEgl5A parent enzymes and the H8 and H9 chimeras. The creation of the fourhomology models used MODELLER 9.19 (56), using PDB structures 5I78 from Aspergillus niger (11), 1H1Nfrom Thermoascus aurantiacus (57), and 5L9C from Penicillium verruculosum as the structural templatesfor system construction. These templates all have greater than 50% sequence identity with each of thefour GH5 sequences considered (Fig. 1), which is typically indicative of strong structural homology (56).Protonation states consistent with the optimal activity pH (5.0) for each GH5 model were determinedusing PROPKA 3.1 (32, 58). Since some of the PDB reference structures feature N-glycosylations (11),which in addition to our direct experimental evidence for glycosylations on our enzymes are commonlyfound in other glycoside hydrolases (2), between 1 and 3 basic N-glycosylations were added to accessibleasparagine residues on each model where appropriate, using the GlyProt Web server (59). Specifically,SoCel5 has glycosylations on Asn23, Asn64, and Asn76. TeEgl5A has four glycosylations, on Asn36,Asn190, Asn219, and Asn267. H8 and H9 each have a single glycosylation at Asn36. Other asparagineresidues were determined to be inaccessible to solution. Each of the four complete GH5 models wassolvated in a water cube with 80-Å sides using the SOLVATE plugin of VMD (60).

Following construction, each of the four GH5 models was simulated for 10 ns with the alpha carbonsharmonically restrained to their initial positions using a 1-kcal mol�1 �2 force constant. This equilibra-tion step permits the placed water molecules and the modelled side chains to find their preferredorientations and rotameric states given the backbone structure. The equilibration was performed in aconstant-pressure and -temperature ensemble, using a Langevin piston (61) as the barostat to maintain1 atm of isotropic pressure and a Langevin thermostat (62) set to 298 K with a 1 ps�1 coupling coefficient.The CHARMM36 (63, 64) force field was used for protein components, together with the CHARMM36carbohydrate force field for the glycosylations (65, 66) and the TIP3 water model (67). Nonbonded termsof the energy function were cut off at 12 Šafter a 10-Šswitching distance. Long-range electrostaticforces were determined using the particle mesh Ewald method (68, 69) with a 1.2-Šgrid spacing.

The end state after this short equilibration was then used as the starting point for all subsequentsimulations. For each model, simulations were carried out at three different temperatures, 298 K (25°C),343 K (70°C, near the melting point), and 398 K (125°C). To minimize the impact of the stochastic natureof these simulations on the final conclusions, each combination of GH5 model and temperature wassimulated five times for 200 ns each. The aggregate 4 microseconds of trajectory were then analyzedusing a Python-enabled build of VMD (60), using its built-in hydrogen bond analysis tools. Atomiccontacts between structural elements were computed using a weighted contacts formula (70, 71) andwere used to identify hydrophobic interactions that are otherwise difficult to quantify.

SUPPLEMENTAL MATERIALSupplemental material for this article may be found at https://doi.org/10.1128/AEM

.02079-18.SUPPLEMENTAL FILE 1, PDF file, 1.4 MB.

ACKNOWLEDGMENTSThis work was supported by the National Natural Science Foundation of China (grant

number 31572446), the Fundamental Research Funds for Central Non-profit ScientificInstitution (grant number Y2017JC31), and the China Modern Agriculture ResearchSystem (grant number CARS-41). G.T.B. thanks the U.S. Department of Energy (DOE)Energy Efficiency and Renewable Energy (EERE) BioEnergy Technologies Office (BETO)for funding under contract number DE-AC36-08GO28308 to the National RenewableEnergy Laboratory (NREL). J.V.V. was supported by the NREL Director’s Fellowship

Zheng et al. Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 16

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

funded by the Laboratory Directed Research and Development (LDRD) program. Thiswork used computing resources provided by the Extreme Science and EngineeringDiscovery Environment (XSEDE) (72), which is supported by National Science Founda-tion grant number ACI-1548562. Specifically, the allocation to G.T.B. (TG-MCB090159)was used on Stampede2, which is hosted by the Texas Advanced Computing Center(TACC) at the University of Texas at Austin.

REFERENCES1. Kuhad RC, Gupta R, Singh A. 2011. Microbial cellulases and their indus-

trial applications. Enzyme Res 2011:280696. https://doi.org/10.4061/2011/280696.

2. Payne CM, Knott BC, Mayes HB, Hansson H, Himmel ME, Sandgren M,Ståhlberg J, Beckham GT. 2015. Fungal cellulases. Chem Rev 115:1308 –1448. https://doi.org/10.1021/cr500351c.

3. Li DC, Li AN, Papageorgiou AC. 2011. Cellulases from thermophilic fungi:recent insights and biotechnological potential. Enzyme Res 2011:308730. https://doi.org/10.4061/2011/308730.

4. Srivastava N, Srivastava M, Mishra PK, Gupta VK, Molina G, Rodriguez-Couto S, Manikanta A, Ramteke PW. 2018. Applications of fungal cellu-lases in biofuel production: advances and limitations. Renew SustainEnergy Rev 82:2379 –2386. https://doi.org/10.1016/j.rser.2017.08.074.

5. Aspeborg H, Coutinho PM, Wang Y, Brumer H, Henrissat B. 2012. Evo-lution, substrate specificity and subfamily classification of glycosidehydrolase family 5 (GH5). BMC Evol Biol 12:186. https://doi.org/10.1186/1471-2148-12-186.

6. Wang K, Luo H, Bai Y, Shi P, Huang H, Xue X, Yao B. 2014. A thermophilicendo-1,4-�-glucanase from Talaromyces emersonii CBS394.64 with broadsubstrate specificity and great application potentials. Appl MicrobiolBiotechnol 98:7051–7060. https://doi.org/10.1007/s00253-014-5680-0.

7. Tseng CW, Ko TP, Guo RT, Huang JW, Wang HC, Huang CH, Cheng YS,Wang AH, Liu JR. 2011. Substrate binding of a GH5 endoglucanasefrom the ruminal fungus Piromyces rhizinflata. Acta Crystallogr Sect FStruct Biol Cryst Commun 67:1189 –1194. https://doi.org/10.1107/S1744309111032428.

8. Lo Leggio L, Larsen S. 2002. The 1.62 Å structure of Thermoascusaurantiacus endoglucanase: completing the structural picture of sub-families in glycoside hydrolase family 5. FEBS Lett 523:103–108. https://doi.org/10.1016/S0014-5793(02)02954-X.

9. Lee TM, Farrow MF, Arnold FH, Mayo SL. 2011. A structural study ofHypocrea jecorina Cel5A. Protein Sci 20:1935–1940. https://doi.org/10.1002/pro.730.

10. Liu G, Li Q, Shang N, Huang JW, Ko TP, Liu W, Zheng Y, Han X, Chen Y,Chen CC, Jin J, Guo RT. 2016. Functional and structural analyses of a1,4-�-endoglucanase from Ganoderma lucidum. Enzyme Microb Technol86:67–74. https://doi.org/10.1016/j.enzmictec.2016.01.013.

11. Yan J, Liu W, Li Y, Lai HL, Zheng Y, Huang JW, Chen CC, Chen Y, Jin J, LiH, Zhong LH, Guo RT. 2016. Functional and structural analysis of Pichiapastoris-expressed Aspergillus niger 1,4-�-endoglucanase. Biochem Bio-phys Res Commun 475:8 –12. https://doi.org/10.1016/j.bbrc.2016.05.012.

12. Liu W, Zhang XZ, Zhang Z, Zhang YH. 2010. Engineering of Clostridiumphytofermentans endoglucanase Cel5A for improved thermostability.Appl Environ Microbiol 76:4914 – 4917. https://doi.org/10.1128/AEM.00958-10.

13. Bayram Akcapinar G, Venturini A, Martelli PL, Casadio R, Sezerman UO.2015. Modulating the thermostability of endoglucanase I fromTrichoderma reesei using computational approaches. Protein Eng Des Sel28:127–135. https://doi.org/10.1093/protein/gzv012.

14. Zhang J, Shi H, Xu L, Zhu X, Li X. 2015. Site-directed mutagenesis of ahyperthermophilic endoglucanase Cel12B from Thermotoga maritimabased on rational design. PLoS One 10:e0141937. https://doi.org/10.1371/journal.pone.0141937.

15. Voigt CA, Martinez C, Wang ZG, Mayo SL, Arnold FH. 2002. Proteinbuilding blocks preserved by recombination. Nat Struct Biol 9:553–558.https://doi.org/10.1038/nsb805.

16. Meyer MM, Hochrein L, Arnold FH. 2006. Structure-guided SCHEMArecombination of distantly related beta-lactamases. Protein Eng Des Sel19:563–570. https://doi.org/10.1093/protein/gzl045.

17. Heinzelman P, Snow CD, Wu I, Nguyen C, Villalobos A, Govindarajan S,Minshull J, Arnold FH. 2009. A family of thermostable fungal cellulases

created by structure-guided recombination. Proc Natl Acad Sci U S A106:5610 –5615. https://doi.org/10.1073/pnas.0901417106.

18. Chang CJ, Lee CC, Chan YT, Trudeau DL, Wu MH, Tsai CH, Yu SM, Ho TH,Wang AH, Hsiao CD, Arnold FH, Chao YC. 2016. Exploring the mechanismresponsible for cellulase thermostability by structure-guided recombi-nation. PLoS One 11:e0147485. https://doi.org/10.1371/journal.pone.0147485.

19. Béguin P. 1999. Hybrid enzymes. Curr Opin Biotechnol 10:336 –340.https://doi.org/10.1016/S0958-1669(99)80061-5.

20. Höcker B, Beismann-Driemeyer S, Hettwer S, Lustig A, Sterner R. 2001.Dissection of a (��) 8-barrel enzyme into two folded halves. Nat StructBiol 8:32–36. https://doi.org/10.1038/83021.

21. Hocker B, Claren J, Sterner R. 2004. Mimicking enzyme evolution bygenerating new (betaalpha)8-barrels from (betaalpha)4-half-barrels.Proc Natl Acad Sci U S A 101:16448 –16453. https://doi.org/10.1073/pnas.0405832101.

22. Smith MA, Rentmeister A, Snow CD, Wu T, Farrow MF, Mingardon F,Arnold FH. 2012. A diverse set of family 48 bacterial glycoside hydrolasecellulases created by structure-guided recombination. FEBS J 279:4453– 4465. https://doi.org/10.1111/febs.12032.

23. Bharat TA, Eisenbeis S, Zeth K, Höcker B. 2008. A ��-barrel built by thecombination of fragments from different folds. Proc Natl Acad Sci U S A105:9942–9947. https://doi.org/10.1073/pnas.0802202105.

24. Sperl JM, Rohweder B, Rajendran C, Sterner R. 2013. Establishing cata-lytic activity on an artificial (��)8-barrel protein designed from identicalhalf-barrels. FEBS Lett 587:2798 –2805. https://doi.org/10.1016/j.febslet.2013.06.022.

25. Huang PS, Feldmeier K, Parmeggiani F, Velasco DAF, Höcker B, Baker D.2016. De novo design of a four-fold symmetric TIM-barrel protein withatomic-level accuracy. Nat Chem Biol 12:29 –34. https://doi.org/10.1038/nchembio.1966.

26. Lombard V, Golaconda Ramulu H, Drula E, Coutinho PM, Henrissat B.2014. The carbohydrate-active enzymes database (CAZy) in 2013. Nu-cleic Acids Res 42:D490 –D495. https://doi.org/10.1093/nar/gkt1178.

27. Richter M, Bosnali M, Carstensen L, Seitz T, Durchschlag H, Blanquart S,Merkl R, Sterner R. 2010. Computational and experimental evidence forthe evolution of a (��)8-barrel protein from an ancestral quarter-barrelstabilised by disulfide bonds. J Mol Biol 398:763–773. https://doi.org/10.1016/j.jmb.2010.03.057.

28. Liu J, Tsai C, Liu J, Cheng K, Cheng C. 2001. The catalytic domain of aPiromyces rhizinflata cellulase expressed in Escherichia coli was stabilizedby the linker peptide of the enzyme. Enzyme Microb Technol 28:582–589. https://doi.org/10.1016/S0141-0229(00)00349-5.

29. Manavalan T, Manavalan A, Thangavelu KP, Heese K. 2015. Character-ization of a novel endoglucanase from Ganoderma lucidum. J BasicMicrobiol 55:761–771. https://doi.org/10.1002/jobm.201400808.

30. Lee KM, Jeya M, Joo AR, Singh R, Kim IW, Lee JK. 2010. Purification andcharacterization of a thermostable endo-�-1,4-glucanase from a novelstrain of Penicillium purpurogenum. Enzyme Microb Technol 46:206 –211.https://doi.org/10.1016/j.enzmictec.2009.11.002.

31. Yu H, Huang H. 2014. Engineering proteins for thermostability throughrigidifying flexible sites. Biotechnol Adv 32:308 –315. https://doi.org/10.1016/j.biotechadv.2013.10.012.

32. Olsson MHM, Søndergaard CR, Rostkowski M, Jensen JH. 2011. PROPKA3:consistent treatment of internal and surface residues in empirical pKapredictions. J Chem Theory Comput 7:525–537. https://doi.org/10.1021/ct100578z.

33. Goedegebuur F, Dankmeyer L, Gualfetti P, Karkehabadi S, Hansson H,Jana S, Huynh V, Kelemen BR, Kruithof P, Larenas EA, Teunissen PJM,Stahlberg J, Payne CM, Mitchinson C, Sandgren M. 2017. Improving thethermal stability of cellobiohydrolase Cel7A from Hypocrea jecorina by

Features Imparting Thermal Stability in GH5 Cellulases Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 17

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from

directed evolution. J Biol Chem 292:17418 –17430. https://doi.org/10.1074/jbc.M117.803270.

34. Sammond DW, Kastelowitz N, Himmel ME, Yin H, Crowley MF, BombleYJ. 2016. Comparing residue clusters from thermophilic and mesophilicenzymes reveals adaptive mechanisms. PLoS One 11:e0145848. https://doi.org/10.1371/journal.pone.0145848.

35. van Dijk E, Hoogeveen A, Abeln S. 2015. The hydrophobic temperaturedependence of amino acids directly calculated from protein structures.PLoS Comput Biol 11:e1004277. https://doi.org/10.1371/journal.pcbi.1004277.

36. Riechmann L, Winter G. 2000. Novel folded protein domains generatedby combinatorial shuffling of polypeptide segments. Proc Natl Acad SciU S A 97:10068 –10073. https://doi.org/10.1073/pnas.170145497.

37. Sharma P, Kaila P, Guptasarma P. 2016. Creation of active TIM barrelenzymes through genetic fusion of half-barrel domain constructs de-rived from two distantly related glycosyl hydrolases. FEBS J 283:4340 – 4356. https://doi.org/10.1111/febs.13927.

38. Zheng F, Huang H, Wang X, Tu T, Liu Q, Meng K, Wang Y, Su X, Xie X, LuoH. 2016. Improvement of the catalytic performance of a Bispora anten-nata cellulase by replacing the N-terminal semi-barrel structure. Biore-sour Technol 218:279 –285. https://doi.org/10.1016/j.biortech.2016.06.094.

39. Lang D, Thoma R, Henn-Sax M, Sterner R, Wilmanns M. 2000. Structuralevidence for evolution of the �/� barrel scaffold by gene duplicationand fusion. Science 289:1546 –1550. https://doi.org/10.1126/science.289.5484.1546.

40. Numata K, Muro M, Akutsu N, Nosoh Y, Yamagishi A, Oshima T. 1995.Thermal stability of chimeric isopropylmalate dehydrogenase genesconstructed from a thermophile and a mesophile. Protein Eng Des Sel8:39 – 43. https://doi.org/10.1093/protein/8.1.39.

41. Hosseini-Mazinani SM, Nakajima E, Ihara Y, Kameyama KZ, Sugimoto K.1996. Recovery of active beta-lactamases from Proteus vulgaris andRTEM-1 hybrid by random mutagenesis by using a dnaQ strain ofEscherichia coli. Antimicrob Agents Chemother 40:2152–2159. https://doi.org/10.1128/AAC.40.9.2152.

42. Shoichet BK, Baase WA, Kuroki R, Matthews BW. 1995. A relationshipbetween protein stability and protein function. Proc Natl Acad Sci U S A92:452– 456. https://doi.org/10.1073/pnas.92.2.452.

43. Conrad B, Hoang V, Polley A, Hofemeister J. 2010. Hybrid Bacillus amy-loliquefaciens X Bacillus licheniformis alpha-amylases. Construction,properties and sequence determinants. Eur J Biochem 230:481– 490.https://doi.org/10.1111/j.1432-1033.1995.tb20586.x.

44. Takano K, Higashi R, Okada J, Mukaiyama A, Tadokoro T, Koga Y, KanayaS. 2009. Proline effect on the thermostability and slow unfolding of ahyperthermophilic protein. J Biochem 145:79 – 85. https://doi.org/10.1093/jb/mvn144.

45. Wang K, Luo H, Tian J, Turunen O, Huang H, Shi P, Hua H, Wang C, WangS, Yao B. 2014. Thermostability improvement of a streptomyces xylanaseby introducing proline and glutamic acid residues. Appl Environ Micro-biol 80:2158 –2165. https://doi.org/10.1128/AEM.03458-13.

46. Shirke AN, Basore D, Butterfoss GL, Bonneau R, Bystroff C, Gross RA.2016. Toward rational thermostabilization of Aspergillus oryzae cutinase:insights into catalytic and structural stability. Proteins 84:60 –72. https://doi.org/10.1002/prot.24955.

47. Shental-Bechor D, Levy Y. 2008. Effect of glycosylation on proteinfolding: a close look at thermodynamic stabilization. Proc Natl Acad SciU S A 105:8256 – 8261. https://doi.org/10.1073/pnas.0801340105.

48. Fonseca-Maldonado R, Vieira DS, Alponti JS, Bonneil E, Thibault P, WardRJ. 2013. Engineering the pattern of protein glycosylation modulates thethermostability of a GH11 xylanase. J Biol Chem 288:25522–25534.https://doi.org/10.1074/jbc.M113.485953.

49. Zhao J, Shi P, Luo H, Yang P, Zhao H, Bai Y, Huang H, Wang H, Yao B.2010. An acidophilic and acid-stable �-mannanase from Phialophora sp.P13 with high mannan hydrolysis activity under simulated gastric con-ditions. J Agric Food Chem 58:3184 –3190. https://doi.org/10.1021/jf904367r.

50. Liu Y-G, Whittier RF. 1995. Thermal asymmetric interlaced PCR: autom-atable amplification and sequencing of insert end fragments from P1and YAC clones for chromosome walking. Genomics 25:674 – 681.https://doi.org/10.1016/0888-7543(95)80010-J.

51. Johnson M, Zaretskaya I, Raytselis Y, Merezhuk Y, McGinnis S, MaddenTL. 2008. NCBI BLAST: a better web interface. Nucleic Acids Res 36:W5–W9. https://doi.org/10.1093/nar/gkn201.

52. Burge C, Karlin S. 1997. Prediction of complete gene structures in humangenomic DNA. J Mol Biol 268:78 –94. https://doi.org/10.1006/jmbi.1997.0951.

53. Bendtsen JD, Nielsen H, von Heijne G, Brunak S. 2004. Improved predic-tion of signal peptides: SignalP 3.0. J Mol Biol 340:783–795. https://doi.org/10.1016/j.jmb.2004.05.028.

54. Miller GL. 1959. Use of dinitrosalicylic acid reagent for determinationof reducing sugar. Anal Chem 31:426 – 428. https://doi.org/10.1021/ac60147a030.

55. Phillips JC, Braun R, Wang W, Gumbart J, Tajkhorshid E, Villa E, ChipotC, Skeel RD, Kalé L, Schulten K. 2005. Scalable molecular dynamicswith NAMD. J Comput Chem 26:1781–1802. https://doi.org/10.1002/jcc.20289.

56. Webb B, Sali A. 2014. Protein structure modeling with MODELLER.Methods Mol Biol 1137:1–15. https://doi.org/10.1007/978-1-4939-0366-5_1.

57. Van Petegem F, Vandenberghe I, Bhat M, Van Beeumen J. 2002. Atomicresolution structure of the major endoglucanase from Thermoascusaurantiacus. Biochem Biophys Res Commun 296:161–166. https://doi.org/10.1016/S0006-291X(02)00775-1.

58. Søndergaard CR, Olsson MH, Rostkowski M, Jensen JH. 2011. Improvedtreatment of ligands and coupling effects in empirical calculation andrationalization of pKa values. J Chem Theory Comput 7:2284 –2295.https://doi.org/10.1021/ct200133y.

59. Bohnelang A, Lieth CWVD. 2005. GlyProt: in silico glycosylation of pro-teins. Nucleic Acids Res 33:W214 –W219. https://doi.org/10.1093/nar/gki385.

60. Humphrey W, Dalke A, Schulten K. 1996. VMD: visual molecular dynamics.J Mol Graph 14:33–38. https://doi.org/10.1016/0263-7855(96)00018-5.

61. Feller SE, Zhang Y, Pastor RW, Brooks BR. 1995. Constant pressuremolecular dynamics simulation: the Langevin piston method. J ChemPhys 103:4613– 4621. https://doi.org/10.1063/1.470648.

62. Kubo R. 1966. The fluctuation-dissipation theorem. Rep Prog Phys 29:255. https://doi.org/10.1088/0034-4885/29/1/306.

63. Best RB, Zhu X, Shim J, Lopes PEM, Mittal J, Feig M, MacKerell AD. 2012.Optimization of the additive CHARMM all-atom protein force field tar-geting improved sampling of the backbonephi, psi and side-chain chi(1)and chi(2) dihedral angles. J Chem Theory Comput 8:3257–3273. https://doi.org/10.1021/ct300400x.

64. Jing H, Rauscher S, Nawrocki G, Ran T, Feig M, de Groot BL, GrubmüllerH, MacKerell AD, Jr. 2017. CHARMM36m: an improved force field forfolded and intrinsically disordered proteins. Nat Methods 14:71–73.https://doi.org/10.1038/nmeth.4067.

65. Guvench O, Hatcher ER, Venable RM, Pastor RW, MacKerell AD. 2009.CHARMM additive all-atom force field for glycosidic linkages betweenhexopyranoses. J Chem Theory Comput 5:2353–2370. https://doi.org/10.1021/ct900242e.

66. Guvench O, Mallajosyula SS, Raman EP, Hatcher E, Vanommeslaeghe K,Foster TJ, Jamison FW, MacKerell AD. 2011. CHARMM additive all-atomforce field for carbohydrate derivatives and its utility in polysaccharideand carbohydrate-protein modeling. J Chem Theory Comput7:3162–3180. https://doi.org/10.1021/ct200328p.

67. Jorgensen WL, Chandrasekhar J, Madura JD, Impey RW, Klein ML. 1983.Comparison of simple potential functions for simulating liquid water. JChem Phys 79:926 –935. https://doi.org/10.1063/1.445869.

68. Essmann U, Perera L, Berkowitz ML, Darden T, Lee H, Pedersen LG. 1995.A smooth particle mesh Ewald method. J Chem Phys 103:8577– 8593.https://doi.org/10.1063/1.470117.

69. Darden T, York D, Pedersen L. 1993. Particle mesh Ewald: an N·log(N)method for Ewald sums in large systems. J Chem Phys 98:10089 –10092.https://doi.org/10.1063/1.464397.

70. Vermaas JV, Petridis L, Qi X, Schulz R, Lindner B, Smith JC. 2015.Mechanism of lignin inhibition of enzymatic biomass deconstruction.Biotechnol Biofuels 8:217. https://doi.org/10.1186/s13068-015-0379-8.

71. Vermaas JV, Tajkhorshid E. 2017. Differential membrane binding me-chanics of synaptotagmin isoforms observed in atomic detail. Biochem-istry 56:281–293. https://doi.org/10.1021/acs.biochem.6b00468.

72. Towns J, Cockerill T, Dahan M, Foster I, Gaither K, Grimshaw A, Hazle-wood V, Lathrop S, Lifka D, Peterson GD, Roskies R, Scott JR, Wilkens-Diehr N. 2014. XSEDE: accelerating scientific discovery. Comput Sci Eng16:62–74. https://doi.org/10.1109/MCSE.2014.80.

Zheng et al. Applied and Environmental Microbiology

March 2019 Volume 85 Issue 5 e02079-18 aem.asm.org 18

on March 1, 2020 by guest

http://aem.asm

.org/D

ownloaded from